AI and data science have a horizontal impact on any industry we can think of, which makes a career in AI and data science exciting, impactful and deeply enriching. When choosing their undergraduate or graduate studies or when looking for a career change, young girls and women need to be made aware of all the opportunities for impact and career paths that AI and data science actually enable. And who is better than other women working in these domains to land this message and provide them with the needed inspiration, mentorship and advice to get there?
Ghida Ibrahim, Quantitative Engineer & Data Scientist, Facebook

Chatting with Ghida, one of the speakers at this evening’s Women in AI Dinner, she highlighted our motivation behind bringing these fantastic women together at events such as these. What stood out most this evening was the collaborative atmosphere and enthusiasm from attendees to share ideas and progressions in their work.

Tonight saw us hosting the 8th London based Women in AI Dinner where we welcomed 4 female keynote speakers and 75 attendees of all genders to join us for an evening of networking and discussions over a champagne reception and a 3 course meal. The evening began with guests discussing their recent work in a variety of areas such as NLP, robotics, healthcare, finance and autonomous vehicles just to name a few, and it was wonderful to see plenty of familiar faces, as well as lots of new attendees.

"I'm not a data scientist but I want to focus more on AI. I've started reading books on statistics to learn more. I'm really interested in bias, data models, and trying to learn the lingo! Last year I came to the dinner and I was a little out of my depth - though it was still valuable! - but this year I think I'll know more! I'm excited to keep having these conversations." - Victoria Hair, Intel

Once we were seated, Chanuki Illushaka Sereshine, AI Researcher/Lead Data Scientist at the Alan Turing Institute/Popsa and our compere for the evening welcomed the buzzing room and reminded us that not only do we need more female role models in AI, but also ‘more visible examples of different types of careers involving AI that might appeal to different types of people.’ She explained that ‘the stereotypical image of a job in this field – coding all day in front of a computer screen and not socialising with anyone – doesn’t appeal to me at all, and fortunately, at least in my life, is far from reality. My day to day job as the lead data scientist at Popsa allows time for lots of creative thinking and collaboration with others, which I really love. This kind of thing really needs to be more visible so we can break the stereotype.’

Chanuki Illushaka Sereshine, Alan Turing Institute 

In her keynote talk focusing on  her primary research at the Alan Turing Institute, Chanuki spoke about ‘Quantifying the Connection Between Scenic Beauty and Our Well Being’. She explained that in our daily lives, it’s natural to seek out beautiful scenery when we want a respite from our busy lives. The topic of AI for a positive impact was a popular one in the conversations that preceded the talks, and she explained that she is using AI to help generate measures of beautiful places, particularly in urban environments, ‘so that I can better understand the connection between the built environment and our wellbeing. This can help us understand how we might design our futures cities to be places we want to live in, as they make us happier.’

She explained that while architects and policymakers have explored this question for centuries, quantitative analyses have been held back by a lack of data. With the ever increasing amount of online data alongside developments in deep learning, there are countless opportunities arising to analyse the beauty of our environment. She is currently using 1.5 million ratings of over 200,000 images covering Great Britain from a website called Scenic-Or-Not to find answers to this age-old question.

Chanuki wanted to find a definition beyond ‘nature is beautiful’. "I was looking at survival theories and other theories. Neural networks are fantastic in computer vision, so I found the network from MIT called 'Places' and I ran the model and found that it’s not just natural places that people were rating highly, they were rating buildings such as cottages, architectural pieces and lighthouses. We have all this data, now what? I have an image for every square kilometre so I though okay let’s throw neural networks at it again.When you’re trying to use neural networks you need millions of samples, but I used transfer learning to overcome this - I took about 200,000 images and trained my neural network to recognise beautiful places around London, and it performed really well on both natural places and places with great architecture."

I love working with neural networks, which I do on a daily basis, as it is absolutely fascinating what we can do with AI and computer vision these days. The rate of progress is astonishing –  there is always something new to learn, which keeps me on my toes and intellectually stimulated.

Chanuki’s work discovered that people report better health and more happiness specifically in more beautiful urban environments. Normally we imagine that only natural environments can help boost our wellbeing, but it is encouraging news to know that beautiful city environments can also play a key role in our wellbeing.

With everyone feeling more positive about their busy city lives and the impact they have on their mental health and well being, Chanuki introduced out next speaker for the evening, Betty Schirrmeister, a Senior Data Scientist at Royal Mail who is currently working on ‘Data Science in Prediction and Delivery Projects.’ One of Royal Mail’s latest initiatives to improve customer experience and convenience was launched in April this year. Customers are now receiving information about their parcel deliveries a day in advance and also get shorter estimated delivery windows, down to a time frame of two hours. This vastly improves customer experience, and to highlight this Betty explained that the estimated delivery window is a good example: ‘using AI, we have improved customer experience providing delivery time windows a day in advance. The project has been launched in October 2018,’ and Betty leads a team of 7 data scientists and engineers to successfully implement this project.  They’ve also been working on finding an end-to-end solution that’s now been running live for two years to support Royal Mail’s daily resource planning. This has resulted in huge financial savings within the company, and this has been deployed via webapp and usage evaluation.

Betty Schirrmeister, Royal Mail

Betty shared that she has always loved science, research, and generally coding. "I think many women are put off programming and stats early on in their careers. Simple, fun coding and data science projects which could be useful in day-to-day life might be an easy start, e.g. building a small game for the kids (Hangman / Connect 4), using ML/AI to predict their monthly expenses, optimising packing for their holiday." Following a charity-hackathon that Royal Mail hosted in 2018, they currently have a data science project in development for ‘Action for Children’, which she hopes will see her work used for a positive social impact.  

Guests enjoyed their starters before moving tables for the next presentation. At every RE•WORK dinner, we try to maximise the number of connections and contacts that everyone makes, so encourage people to move to a different table between courses. Once the plates were cleared, we were back on track and hungry to learn from our next speaker. Next up to present was Maren Eckhoff from QuantumBlack, our fantastic sponsor for the evening. Maren, Principal Data Scientist shared her work on ‘Solving real-world black box optimisation’. Maren has a Maths background and PhD in Probability, and began by explaining that ‘optimisation problems are ubiquitous in the field of AI.’ Throughout her talk, Maren focused on the optimisation of operational decisions to maximise production. She explored modern black box optimisation techniques like Bayesian Optimisation and Genetic algorithms as well as the value of thinking about complex systems as graphs. "As Data Scientists, we are trained to think in tables. A graphical representation often offers new avenues to a problem and leads to interpretable solutions."

Maren Eckhoff, QuantumBlack
We’ve been working with a major player in the oil and gas industry on a complex optimisation problem. Due to the highly non-stationary environment, our solution had to rely on simulation instead of historic data. The high dimensionality of the problem and unknown, complex objective function made this a difficult challenge. One of the keys to solving the problem was a close collaboration with domain experts. Knowledge sharing between different disciplines allowed us to reduce the solution space and guarantee realistic recommendations. Without bringing domain expertise and Data Science, we could not have solved this problem.

QuantumBlack also sponsored the Boston edition of the Women in AI Dinner, and Alison O’Connor joined us in an interview, sheding light on a ‘day in the life’ of a female expert in the field, which you can read here.  

It was then time for the main course and many more discussions were flowing as everyone enjoyed their food and wine. Once again, attendees were encouraged to move seats, Ghida Ibrahim was the final expert to present. The Quantitative Engineer & Data Scientist from Facebook spoke on human/machine interaction, discussing ‘How Data Analytics Can Help Us Build A More Empathetic Internet?’ She began by explaining that ‘as much as having access to water, electricity, shelter and food is a basic human right, in the 21st century, having internet access is increasingly becoming a human right too.’ People from across the globe from different backgrounds and genders and with different jobs and interests all use the internet differently, ‘and for many of them the internet presents a tool for economic and social empowerment. That is why I think that internet should be accessible to everyone, and that everyone should have an enjoyable internet experience tailored to their specific interests and needs.’

Ghida Ibrahim, Facebook

Ghida explained that she builds data driven tools and models, and performs in depth analysis to drive the expansion and optimise the operation of one of the largest and most complex networks forming the internet, with the goal of bringing more people online to a better internet.

"I've spent the last 8 years of my career working in the internet and I still don't understand it fully! Overall, we want everyone to be able to have the best experience from the internet. Many people think the internet is a black box. It's actual definition is a "Network of interconnected networks". By definition it means that it is operated by different players with different geographic footprints."

When asked what her favourite aspect of her job role is, Ghida replied: ‘First, working on challenging problems that matter. Second, the scale of the impact. Third and most importantly, the people, it is great to work on a daily basis with people that are smarter than you yet so humble and eager to learn.’

In her spare time, Ghida built Rafiqi, a platform that leverages artificial intelligence (AI) for connecting refugees to life opportunities, that has been recognized by TechCrunch as one of the most innovative new projects using tech to help refugees, and was a finalist of many awards including Techfugees Global Challenges competition and the Tech for Good UK awards.

Maren, Betty, Chanuki and Ghida

With the final presentation drawing to a close, there were plenty of discussions surrounding the positive impacts of AI on society, as well as some of the algorithms and models that the speakers and attendees alike were using. Over dessert, we had the chance to chat to some of the guests and hear what they thought of the evening:

• "I went to the Deep Learning Summit 3 years ago in 2016 when I finished my PhD. I was in the transition of leaving academia and heading into industry and I wanted to see how my research could be applied. Now I'm back and ready to learn more and see how the industry has moved on." - Laura Acqualagna, Technische Universität Berlin

"So we’ve got Cars, Healthcare, it’s already really diverse. It also seems really laid back but with some great conversation opportunities, which is exactly why I came." - Klaudia Ludwisiak, McLaren Applied Technologies

• "Not many people about our company but there are lots of brands that work under one umbrella. We’re working on applied AI. We have a lot of data so there’s lots of ways we can help industry. It’s relevant across loads of devices." - Anon.

Did you miss out on the Women in AI Dinner this time? Join us at the next edition of the London dinner this November by joining the waiting list now, and you’ll be the first to know when tickets go on sale. In the meantime, listen to the new season of the Women in AI Podcast focusing on Social Good, where we'll hear from our first guest of the season Alice Xiang from the Partnership on AI, discussing responsibility, fairness and transparency in AI.